Elders with bipolar disorders (BP) have severe and debilitating psychopathology, poor treatment outcomes, and high medical morbidity and mortality. Controlled treatment studies for late-life BP disorder are lacking, and findings in young patients may not be relevant because aging-related changes may reduce the tolerability and efficacy of treatments. Improving the treatment of late-life BP disorder was identified as a priority at the NIMH-DBPA Consensus Conference (2001). In response, we propose the first randomized, double-blind, controlled comparison of lithium (LI) and divalproex (DV) in elders with BP-I manic, mixed and hypomanic episodes, designed to test differences in side effects and benefits under dosing conditions relevant to geriatric practice. Hospitalized and ambulatory patients (n=306) aged 60 years and older will be studied for 9 weeks. Participants will be randomized to LI (0.4-0.99 mEq/L;target 0.80-0.99 mEq/L) or DV (valproate 40-99 meg/ml;target 80-99 meg/ml), and will be treated with monotherapy in the first three weeks at maximum tolerated concentrations. Patients with unsatisfactory response after three weeks will receive adjunctive olanzapine. The tolerability measures will be side effect ratings and the proportion treated at target concentrations. The efficacy measures will be improvement in mood symptoms and in function and quality of life. Data obtained for testing the primary hypotheses will allow secondary analyses of selected factors which may modify outcomes. Compatibility of selected measures with those of the STEP-BD study of younger patients will enhance the value of findings from this study. Recognizing the difficulties inherent in a randomized study of BP elders, we brought together six groups (Cornell, Baylor College of Medicine, Case Western Reserve, Duke, and Universities of Pennsylvania and Pittsburgh) with research expertise in late-life and young adult BP disorders, and with strong research infrastructure and clinical resources. The sites have a history of successful collaborative research, including demanding geriatric studies.